The spatial variability of overbank sedimentation on river floodplains

Abstract

River floodplains are important landforms, and information on rates and patterns of overbank sediment deposition is essential for understanding their development. Such information is also needed for validating existing overbank floodplain flow and sediment transport and deposition models. The use of bomb-derived ¹³⁷Cs for documenting rates of overbank sediment deposition on river floodplains offers the potential for assembling data for a large number of points on a floodplain and therefore for investigating the spatial patterns involved. This approach also overcomes some of the operational and sampling problems encountered in traditional monitoring techniques. The study reported uses ¹³⁷Cs measurements on sediment cores to investigate the spatial variability of sediment deposition on selected floodplain reaches of five British lowland rivers. Attention is given both to the overall reach and to more detailed investigations of small areas. The observed spatial patterns of sediment deposition are interpreted in terms of the major controls involved, which include the floodplain microtopography and morphology. The relationship between the grain size composition of deposited sediment and sedimentation rate is also considered.

Introduction

River floodplains have attracted increasing attention in recent years (cf. Anderson et al., 1996). This interest reflects, at least in part, the ecological significance of floodplains and river corridors within the landscape, their importance as a buffer between the river and the surrounding land, and their role in providing storage for flood control. There is also a growing awareness of their potential significance as sinks for river-borne sediment and associated nutrients and contaminants. Because of their dynamic nature, sediment deposited on river floodplains may be reworked in the future and may thus also constitute a problem for future river management (cf. Leenaers and Schouten, 1989). Against this background, there is clearly a need for an improved understanding of the geomorphological evolution of floodplains and their role as sediment sinks.

When considering the evolution of floodplains and their role as sediment sinks, attention commonly focuses either on the coarse channel deposits and the interaction between channel migration and floodplain construction and destruction (cf. Wolman and Leopold, 1957; Howard, 1992) or on the fine overbank deposits which mantle the large areas of most floodplains and result in vertical accretion of the floodplain surface. For most lowland rivers, particularly those where channelisation and river training works constrain channel migration, overbank deposition of fine sediment will represent the dominant component of floodplain development and evolution. In order to develop an improved understanding of this component of floodplain development, more information on the magnitude and spatial variability of overbank sedimentation rates is required both to quantify the processes involved and to provide a basis for model development (cf. Pizzuto, 1987; Nicholas and Walling, 1996, Nicholas and Walling, 1997).

Existing approaches to documenting rates of overbank sedimentation on river floodplains have included the use of sedimentation traps (e.g., Gretener and Strömquist, 1987; Lambert and Walling, 1987; Asselmann and Middelkoop, 1995), post-event surveys of the deposits resulting from individual floods (e.g., Brown, 1983; Marriott, 1992; Walling et al., 1997), the identification of datable levels within the overbank deposits (e.g., Costa, 1975; Trimble, 1983; Hupp, 1988; Lewin and Macklin, 1987) and the use of maps or ground surveys to establish the maximum ages of observed fine-unit thicknesses in situations where channels are laterally mobile (e.g., Lewin, 1987). The first two approaches are primarily applicable to individual events, although results obtained could be summed to provide data for longer periods. However, the use of sedimentation traps faces problems of representativeness and the need to deploy the traps in advance of the flood event, as well as practical constraints on sampling density, whilst post-event surveys are heavily reliant upon the existence of measurable depths of sediment and of a clearly defined interface between the `new' sediment and the original surface. The use of datable levels or surfaces to estimate sedimentation rates will by definition provide longer-term average values which will integrate inter-event variability conditioned by such factors as flood magnitude and duration and suspended sediment concentrations. However, the viability of this approach is ultimately entirely dependent upon the ability to establish datable levels, and the lack of temporal resolution may prove problematical where long timescales and changing sedimentation rates are involved. The use of maps or ground surveys is limited to areas where channels are laterally mobile and also requires detailed information on changes of floodplain morphology and topography over a certain period of time, which may not be available for the floodplain sites of interest.

Recent advances in the application of environmental radionuclides, and more particularly the fallout radionuclide caesium-137 (¹³⁷Cs) and unsupported lead-210 (²¹⁰Pb), to documenting rates and patterns of floodplain sedimentation can be viewed as an extension of this latter approach, wherein down-profile variations in the concentrations of these radionuclides can provide a basis for establishing the recent chronology of overbank sediment deposition, and measurements of total radionuclide inventories can be used to estimate average sedimentation rates (cf. Walling and He, 1992, Walling and He, 1994, Walling and He, 1997; He and Walling, 1996a, He and Walling, 1996b; Walling et al., 1996). Such use of fallout radionuclides affords many advantages over other methods for documenting overbank floodplain sedimentation, including the general applicability of the approach to a wide range of environments and locations, the medium-term timescales involved (i.e., ca. 40 years for ¹³⁷Cs and ca. 100 years for unsupported ²¹⁰Pb) and the potential for assembling data for a large number of points on a floodplain and thereby for investigating the spatial variability of deposition rates and the patterns involved.

¹³⁷Cs is an artificial fallout radionuclide with a half-life of 30.17 years which was introduced into the environment by the atmospheric testing of thermonuclear weapons primarily during the late 1950s and the 1960s. The depth distribution of the radionuclide in floodplain sediments will reflect the temporal pattern of fallout and the peak ¹³⁷Cs activity can thus be ascribed to 1963, the time of maximum fallout. Dating of this level provides a means of estimating the average sedimentation rate over the past ca. 30–35 years (cf. Walling and He, 1992, Walling and He, 1997; He and Walling, 1996a). Comparison of the total ¹³⁷Cs inventory for a floodplain core with those of adjacent natural undisturbed soils, in order to establish the excess inventory associated with sediment deposition, provides an alternative means of estimating the average sedimentation rate over the period since the onset of significant radiocaesium fallout (i.e., since ca. 1954). In the latter case, only a single measurement of the total inventory of the bulk core is required, rather than measurements of the individual sections of the core needed to define the depth distribution of ¹³⁷Cs, although it is necessary to take into account variations in grain size composition between individual floodplain cores (cf. Walling et al., 1996; Walling and He, 1997). Because of the time consuming nature of ¹³⁷Cs measurements, use of whole core inventory values to estimate sedimentation rates provides greater scope for assembling information for a large number of points on the floodplain, as would be required to investigate spatial patterns of sedimentation, and which are needed to validate existing overbank floodplain flow and sediment transport and deposition models (cf. James, 1985; Pizzuto, 1987; Howard, 1992, Howard, 1996).

The study reported here aimed to use fallout ¹³⁷Cs measurements of floodplain sediment cores to document rates and spatial patterns of overbank sediment deposition on the floodplains of the middle or lower courses of five British lowland rivers. These include the River Stour in Dorset, the River Culm in Devon, the River Severn in Shropshire, the River Rother in West Sussex and the River Avon in Warwickshire. To study the local variability of sediment deposition on these floodplains at different spatial scales, both general down-reach trends and small-scale spatial patterns were documented for each floodplain reach investigated. Particular attention was given to using the observed spatial patterns to elucidate the influence of contrasts such as local floodplain microtopography and morphology on deposition rates.